Arxiv:1805.08185V2 [Physics.Atom-Ph] 21 Jan 2021 Δω Δα Δµ = Qα + Qµ

Enhanced sensitivity to ultralight bosonic dark matter in the spectra of the linear radical SrOH

Ivan Kozyryev,1, 2, ∗ Zack Lasner,1, 2,† and John M. Doyle1, 2 1Harvard-MIT Center for Ultracold Atoms, Cambridge, MA 02138 2Department of Physics, Harvard University, Cambridge, MA 02138 (Dated: January 22, 2021) Coupling between Standard Model particles and theoretically well-motivated ultralight dark matter (UDM) candidates can lead to time variation of fundamental constants, including the proton-to-electron mass ratio µ ≡ mp/me ≈ 1836. The presence of nearly-degenerate vibrational energy levels of different character in polyatomic molecules can result in significantly enhanced relative energy shifts in molecular spectra originating from ∂t µ, relaxing experimental complexity required for high-sensitivity measurements. We analyze the am- plification of UDM effects in the spectrum of laser-cooled strontium monohydroxide (SrOH). SrOH was the first polyatomic molecule to be directly laser cooled to sub-millikelvin temperatures [Kozyryev et al., Phys. Rev. Lett. 118, 173201 (2017)], opening the possibility of long experimental coherence times and providing a 3 promising platform for suppressing systematic errors. Because of the high enhancement factors ( Qµ ≈ 10 ), measurements of the X˜ (200) ↔ X˜ 0310 rovibrational transitions of SrOH in the microwave regime can result in ∼ 10−17 fractional uncertainty in δµ/µ with one day of integration, leading to significantly improved constraints for UDM coupling constants. We also detail how the use of more complex MOR-type radicals with additional vibrational modes arising from larger ligands R could lead to even greater enhancement factors, while still being susceptible to direct laser cooling.

I. INTRODUCTION sensitivity [26–28]. Additionally, specific atomic transitions with enhanced sensitivities |Qα| 1 have allowed measure- The quantum mechanical nature of dark matter remains a ments on a Dy beam to be competitive with atomic clock mystery despite significant experimental efforts [1–5]. Strin- limits [13, 29]. Exploring both Γµ and Γα is important as gent limits placed recently on the promising class of dark mat- these effects probe different underlying physical phenomena ter candidates, Weakly Interacting Massive Particles [3,6], as [18]. While the use of atomic clocks for probing dark-matter- well as the absence of signatures for supersymmetric partners induced oscillating, drifting and transient-in-time fundamen- at the Large Hadron Collider [1,7,8] and electron electric tal constants has been considered in depth [21, 30, 31], laser- dipole moment (EDM) experiments [9, 10], have motivated a cooled molecules have additional degrees of freedom that new generation of searches for other theoretically motivated could enable further breakthroughs in this area. dark matter candidates [11–17]. Bosonic ultralight dark mat- In molecular spectra, the energy scales for electronic, vibra- −1/2 ter (UDM) particles, like axions, axion-like particles (ALPs), tional and rotational transitions typically relate as 1 : µ : −1 dilatons, moduli, and relaxions [11], can form coherently os- µ [32]. Molecular transitions provide a system to study Γµ cillating classical fields φ(r, t) = φ cosω t − k ·r with the couplings without any contributions from Γα because vibra- 0 φ φ 1 oscillation frequency set by the mass of the dark matter par- tional transitions in molecules have Qµ = − 2 and Qα ≈ 0 µ ticle ωφ ' mφ [18–20]. Coupling between UDM fields and [23]. Thus, isolating effects from variation in a model- ordinary matter can lead to variation in fundamental constants independent manner becomes possible [33]. Moreover, cer- X = α (fine-structure constant) and µ (proton-to-electron mass tain beyond the Standard Model theories predict larger µ vari- ratio) as [19, 21] ation δµ/µ = Rδα/α with R ≈ 40 [23, 34, 35], further motivat- ing precision experiments in molecular spectroscopy. Molec- δX (t) n = ΓX φ (r,t), (1) ular ions can also be used for such experiments and recent the- X oretical proposals consider using diatomic hetero- and homo- where the coupling strength is ΓX and n = 1(2) for lin- nuclear molecular ions to search for µ variation [36, 37]. In ear(quadratic) coupling. this paper, we propose to use laser-cooled samples of the neu- Transitions between different quantum levels with energy tral polyatomic radical SrOH that can be trapped at high den- separation 4E = ~ω in atoms and molecules are dependent sities and low temperatures, allowing for large scalability and on the dimensionless constants δω = f (δα,δµ) with [22, 23] enhanced sensitivity to UDM-induced µ variation. arXiv:1805.08185v2 [physics.atom-ph] 21 Jan 2021 δω δα δµ = Qα + Qµ . (2) ω α µ II. ENHANCED SENSITIVITY TO DARK MATTER WITH Sensitive probes of α variation due to UDM-induced ef- NEAR-DEGENERATE STATES fects have recently been explored with the use of ultrapre- −17 cise atomic clocks [18, 24, 25], reaching ∂t α/α ∼ 10 /yr As previously pointed out [22, 38–41], rovibrational spectroscopy of diatomic and polyatomic molecules may provide significant enhancements in relative sensitivity to the variation ∗ [email protected] of µ with Qµ 1. An extensive list of enhancement factors † [email protected] calculated for diatomic and polyatomic molecules to µ varia- 2

To see how a large sensitivity to µ variation arises in the rovibrational spectrum of SrOH, we begin by following previous treatments in Refs. [22, 23]. Consider two different energy levels Eg and Ee within the same electronic state with Eg < Ee. The energy difference is (assuming ~ = 1)

ω = Ee − Eg (3) (0330) with the change arising from µ variation given as (200) ∂E ∂E (0310) δω = e δµ − g δµ. (4) (0220) ∂µ ∂µ (100) (0200) Therefore, the fractional change in the level separation is (0110) δω 1 ∂Ee ∂Eg -1 = − δµ. (5) 527 cm ω Ee − Eg ∂µ ∂µ 364 cm-1 ~ Equivalently, the relationship between the fractional changes X2Σ+ (000) in ω and µ are related to each other as δω δµ Figure 1. Vibrational levels of SrOH in the electronic ground state. = Qµ (6) The dominant nature of the vibrational motion is indicated schemat- ω µ l ically at the top. The states are labeled using v1v2v3 notation for with the proportionality constant Qµ also known as the dimen- the number of vibrational quanta in the Sr-O stretching (v1), Sr-O-H sionless enhancement factor defined as bending (v2) and O-H stretching (v3) vibrational motion. The number of units of vibrational angular momentum present in the doubly- µ ∂Ee ∂Eg Qµ ≡ − (7) degenerate bending of linear triatomic molecules is denoted with the Ee − Eg ∂µ ∂µ superscript l. Nearly-degenerate excited stretching (200) and bend- or as more common in the literature ing 0310 vibrations are indicated with a red oval. 1 ∂E ∂E Q ≡ e − g . (8) µ ω ∂(lnµ) ∂(lnµ) tion can be found in Refs. [22, 23], with large enhancements The absolute dependence of each energy level is calculated as of Q ∼ 300 and Q ∼ 700 estimated for CH OH and l-C H, µ µ 3 3 ∂Eg,e respectively. Other polyatomic molecules found in space like qg,e ≡ (9) ∂(lnµ) methanol [42], acetone [43] and ammonia [44] have been ana- −7 lyzed as well, leading to a stringent limit of δµ/µ . 10 from and has units of energy (usually cm-1). From Eq.8 one can the observations of astronomical methanol [23, 45]. While as- observe that a large enhancement factor Qµ will arise when trophysical observations place stringent time-variation limits two levels being probed are closely spaced (i.e. ω ≈ 0) and −17 with ∂t µ/µ ∼ 10 /yr bounds due to large look-back times have different dependence on µ (i.e. qg 6= qe). (4t ∼ 7Gyr) [23], they have limited sensitivity to UDM- Generically, the interplay between harmonic and anhar- induced coherent oscillations since a linear drift, ∂t µ = δµ/4t, monic contributions (discussed in detail for SrOH below) must be assumed. to the difference in sensitivity coefficients, 4q, can lead to Here we analyze the enhancement factors for one of the enhancement factors Qµ significantly larger than unity. In simplest possible polyatomic molecules, the linear triatomic order to demonstrate the role of both harmonic and anhar- XYZ-type radical SrOH, and discover that enhancement fac- monic terms, we consider two vibrational levels separated 3 tors of Qµ ≈ 10−10 can be reached by probing rovibrational by 4E = 4Eharm + 4Eanharm. Using the dependence of vi- ˜ ˜ 1 brational constants on the proton-to-electron mass ratio (see transitions of the X (200) ↔ X 03 0 excitation spectrum in 1 the ω ≈ 2π × 1 − 30 GHz transition frequency band. SrOH App.C) we calculate 4q = − 2 4Eharm − 4Eanharm for the µ was the first polyatomic molecule to be directly laser cooled sensitivity difference. Therefore, the absolute enhancement [46], and thus provides the additional significant advantages factor Qµ = 4q/4E becomes of low translational and internal temperatures, long experi- 1 ∆E Q = − 1 + anharm . (10) mental coherence times, and optical internal state prepara- µ 2 4E + 4E tion and efficient readout. Furthermore, the simple vibrational harm anharm structure of SrOH strongly limits the possibility of internal vi- For illustration, we consider three limiting cases, depending brational redistribution (IVR) or nonradiative transitions [47], on the relative contributions of 4Eharm and 4Eanharm: enabling highly sensitive laboratory measurements of both 1 4E = 0 : Q → − δµ/µ and ∂t µ/µ in the frequency band of theoretical interest anharm µ 2 for promising UDM models. Figure1 shows the relevant vi- 4E = 0 : Q → −1 (11) brational energy levels of SrOH in the ground electronic state harm µ X˜ . 4Eharm ∼ −4Eanharm : Qµ → ±∞. 3

2 Therefore, a large enhancement factor is expected for a transi- −giili − B[v]N (N + 1), tion with anharmonic contributions comparable in magnitude to the harmonic oscillator energy difference and opposite in where l1(l2) = 0(1), d1(d2) = 1(2) and the sensitivity q[0,000] sign. Inclusion of the small rotational energy difference 4Erot of the ground vibrational level has been subtracted. Each of in a given rovibrational transition leaves Eq. 10 unchanged up the rotational levels in the 0310 vibrational state consists to the substitution 4Eanharm → 4Eanharm + 4Erot. 2 of `-type parity doublets separated by ∆E±l ∼ O B /ω2 We now consider the harmonic and anharmonic contribu- [v] which has been measured for SrOH in this specific vibrations to the energy of rovibrational states in SrOH. As shown tional level to be ∆E ≈ 12 MHz [51]. Driving the per- in App.A, for a linear triatomic molecule like SrOH, the posi- ±l pendicular vibronic transition Σ − Π with 4l = ±1 leads to tions of the vibrational energy levels v vl v referenced rel- 1 2 3 P and R branches with 4N = ±1, as well as a strong Q ative to the lowest level (000) in a given electronic state are branch with 4N = 0 [52]. The relative sensitivity coeffi- described as [48] cient of the rovibrational N00 = 1 → N0 = 1 transition for 1 3 X˜ (200) → X˜ 03 0 is estimated to be Qµ = −617 with tran- 2 2 00 Ev v v −000 = ωivi + xiivi + xiividi + g22l2 (12) sition frequency ω = 2π × 1.1 GHz. By choosing the N = 1 2 3 ∑ 0 i=1 2 → N = 1 rotational branch instead, we obtain Qµ = −23 with ω = 2π × 31 GHz. The sign of the shift can be reversed with di = 1 for the stretching modes with frequencies ω1 by using the other transition branch N00 = 1 → N0 = 2 with ↔ ↔ d = (Sr O) and ω3 (O H), and i 2 for the doubly-degenerate Qµ = 23 and ω = 2π × 29 GHz. Thus, by measuring dif- bending mode with frequency ω2. The anharmonic contri- ferent rotational branches of the same vibrational transition butions to the molecular potentials have been included lead- the sign and magnitude of the sensitivity enhancement fac- x g ing to additional ii and ii terms in the expansion. The ex- tor Qµ can be controlled. Vibrational dependence of the rota- pressions for the two closely-lying vibrational levels of SrOH tional constant B[v] can be used to achieve even larger Qµ since shown in Fig.1 are given as E200−000 = 2ω1 + 6x11 and 00 0 4B200−0310 ≈ −45 MHz [49, 51]. For the N = 5 → N = 5 E0310−000 = 3ω2 + 15x22 + g22. With the estimated molecu- rotational branch, the separation between the levels is esti- lar constants for SrOH based on experimental measurements mated to decrease to 4E200−0310 < 200 MHz, resulting in [49], we determine the energy separation between the two 3 Qµ > 10 enhancement (see Sec.IIA for discussion of the states 4E 1 to be 200−03 0 uncertainty in these estimates). As a stability reference, one could use purely rotational transitions within the (200) vibra- 2ω + 6x − 3ω − 15x − g = 0.0395cm−1, (13) 1 11 2 22 22 tional manifold with Qµ = −1. It is important to note that our spectroscopic constants derived from previous experimental which corresponds to about 1.2 GHz. As discussed measurements [49] reproduce positions of E100, E200, E0110, above, because the harmonic and anharmonic contributions -1 E0200 and E0220 to within 0.002 cm (see App.A). Further- to ∆E200−0310 = ∆Eharm + ∆Eanharm depend differently on more, the absolute magnitude of the calculated enhancement µ, the transition frequency displays a strong sensitivity to factors Qµ is comparable to the largest values found in the µ that is not suppressed even in the limit of degener- literature for much more complex polyatomic molecules like acy. In this regime, extremely small absolute energy shifts, methanol [53] and ammonia [44]. δ∆E200−0310 < 10µHz, can be experimentally resolved, providing a sensitive probe of δµ ∝ δ∆E200−0310. While the dominant energy scale arises from vibration, the A. Enhancement factor uncertainty smaller contribution to Qµ from rotational motion becomes important when the vibrational energies between two states are nearly degenerate. For the ground electronic state X˜ of In our analysis of the anharmonic contributions to the vi- SrOH, the valence electron is effectively localized on the Sr brational potential of SrOH, we have ignored the terms aris- atom and the unpaired electron spin is not strongly bound to ing from coupling between different vibrational modes (i.e. the internuclear molecular symmetry axis z [50]. Therefore, xi j with i 6= j) in Eq. A1. While the vibrational potential rotational levels in both (200) and 0310 vibrational states for SrOH is mostly harmonic with ωi xii, xi j, contributions can be analyzed in terms of Hund’s coupling case (b) quan- from the xi j terms could lead to shifts on the order of a few -1 tum numbers [48] as F (N) = B N (N + 1), where N is the cm . Previous experimental bounds on the location of the [v] [v] 3 quantum number of the total angular momentum apart from 03 0 vibrational level along with the estimate of the g22 co- -1 spin and B[v] is a rotational constant for a specific vibrational efficient further confirm that 4E0310−200 . 2 cm [49]. While level [v]. exact spectroscopy of the 0310 level in reference to a known Using the dependence of the harmonic (ωi), anharmonic vibronic level is necessary to determine the separation be- 1 (xii, gii) and rotational (Bi) coefficients on the proton-to- tween (200) and 03 0 , we estimate an absolute worst-case electron mass ratio µ [23], we calculate the absolute sensitivity value of |Qµ| ≈ 100. For a generic value of 4E0310−200 < 2 of each rovibrational level [N,v] to be cm-1, we can identify a new optimal pair of rotational levels to use in a P or R branch as ∂E[N,v] 1 2 q[N,v] ≡ = − ωiv − xii v + vdi (14) ∂(lnµ) 2 4E0310−200 ≈ −∆Erot = B(N (N + 1) − (N − 1)N) (15) 4

−1 4E 1 2cm perimental run. Long coherence times with laser-cooled SrOH ⇒ N ≈ 03 0−200 ≤ = 4 (16) 2B 2 × 0.25 cm−1 molecules can be realized utilizing either an optical dipole trap or a molecular fountain [62, 63]. The required experimental where ∆Erot is the difference in rotational energies and coherence time Tc is a factor of 5 shorter than the achieved we used for the rotational constants B ≡ B ≈ B 1 ≈ 200 03 0 lifetime of laser-cooled CaF in a red-detuned optical dipole 0.25cm−1 [49, 54]. In the worst case, the total angular tran- trap [62]. Alternatively, a blue-detuned “box” trap [64] would sition frequency ω = |∆E 1 + ∆E | cannot be made 03 0−200 rot enable similarly long trap times. Precision spectroscopy of smaller than B. Therefore, ω < 2π×7.5 GHz and |Q | > 100. µ laser-cooled atomic radium has previously been performed in In a typical (rather than worst-case) scenario, enhancement an optical dipole trap [65], demonstrating the feasibility of the factors significantly larger than this limit would be achieved. optical approach. Thus, comparable sensitivity can be reached as estimated us- 6 ing the best-fit spectroscopic constants currently available. With 10 trapped molecules per experimental cycle, repeated every Tc, and one day of experimental integration, an absolute statistical uncertainty of δω ≈ 10µHz can be III. SENSITIVITY ESTIMATION achieved. Enhanced sensitivity coefficients Qµ in SrOH spectra provide an opportunity to perform sensitive measurements with relaxed experimental precision, similar to gains In addition to the large relative enhancement factors to µ in α variation sensitivity for Dy experiments [29]. The fre- value variation, SrOH uniquely provides an intriguing ex- quency of the rotational transitions addressed during the experimental platform for achieving precise measurements of periment on the X˜ (200) ↔ X˜ 0310 vibrational band ranges δµ/µ using previously demonstrated atomic physics technolo- between 1 and 30 GHz, and therefore expected relative mea- gies. For atomic clock experiments, the statistical precision surement uncertainty is between 3×10−12/pτ(seconds) and with which the transition frequency can be measured, with −13 p the frequency stability limited by quantum projection noise, is 1 × 10 / τ(seconds). For comparison, microwave fre- √ quency synthesizers in the comparable frequency range ω ∼ δω ≈ 1/ NTcτ [36, 55], where N is the number of indepen- T 2π × 10GHz for use in atomic clock experiments have micro- dent molecules probed per run, c is the experimental coher- −14 p ence time, and τ is the total measurement time. Vibrational hertz resolution and noise levels at the 10 / τ(seconds) motions of SrOH are quite harmonic for low quantum num- level [66]. bers and, therefore, radiative vibrational decays with 4v 6= ±1 Combining this expected frequency precision with the en- are suppressed. Thus, the coherence time in the experiment hancement factors estimated in Sec.II, we can achieve a frac- µ/µ ∼ × −17 Tc will be limited by the spontaneous vibrational decay from tional sensitivity δ on the order of 1 10 for both X˜ (200) to X˜ (100), which we estimate to be ∼ 140 ms (see ω ≈ 2π × 1 and 2π × 30 GHz transition frequencies (for a de- Sec.IVC). Black-body stimulated lifetime at room tempera- tailed discussion of the frequency-dependent sensitivity under different measurement scenarios, see Sec.VI). Thus, mi- ture is estimated to be TBBR > 1.5 s, consistent with previous theoretical estimates [56]. crowave spectroscopy of SrOH can provide δµ/µ sensitivity Exploiting the full coherence time of the X˜ (200)−X˜ (0310) at the level of the best previously proposed ultracold atom and system requires laser cooling and trapping SrOH molecules. trapped diatomic neutral [38–40] and ionic species [36, 37], Direct laser cooling of 106 SrOH molecules to millikelvin but with potentially easier experimental preparation and spec- temperatures has already been demonstrated [46]. With troscopy schemes, as well as suppression of systematic errors the Doppler cooling technique, which relies on the sponta- as described in Sec.V. Furthermore, the measurement with neous radiation pressure force, the transverse temperature of SrOH would lead to orders of magnitude improvement in the a cryogenic SrOH beam was reduced to 30 mK [57]. Addi- limit on µ variation in a model independent way compared tionally, the use of the sub-Doppler cooling method known to the previous experimental results with SF6 beam spec- as magnetically-assisted Sisyphus laser cooling reduced the troscopy [33] or photoassociated ultracold KRb molecules −14 temperature to ∼ 750µK[46]. Detailed measurements of where . 10 /yr sensitivity was achieved [67]. Franck-Condon factors (FCFs) and vibrational branching ra- tios (VBRs) for SrOH have been completed [58], confirming that direct laser slowing and magneto-optical trapping appears IV. EXPERIMENTAL DETAILS feasible predominantly with three repumping lasers to address 0 losses to the (100), (200), and 02 0 states. Potentially, In this section, we show in greater detail the feasibility of even fewer repumping lasers could be used employing slow- transferring population to one of the nearly-degenerate vibra- ing with coherent stimulated optical forces recently experi- tional states, driving the nominally forbidden microwave tran- mentally demonstrated for SrOH [59]. Sympathetic cooling sition, and achieving long vibrational coherence times. of trapped SrOH to microkelvin temperatures with ultracold lithium also appears feasible based on rigorous quantum scat- tering calculations [60]. Direct magneto-optical trapping of ∼ 106 diatomic CaF molecules has already been demonstrated A. State preparation [61]. With a combination of these demonstrated techniques, it is The efficient preparation of the necessary rovibrational realistic to assume N ≈ 106 trapped SrOH molecules per ex- quantum states can be achieved in two distinct ways. First, 5

2 moment function lead to overtones of reasonable intensity Π3/2 with 4v = ±2, ±3,... [52]. Additionally, for polyatomic 2 2Π Π3/2 1/2 molecules with nearby vibrational levels of different sym- 2 metry character (e.g. Σ vs Π) like SrOH, Coriolis perturba- 2 Π1/2 ~ Π3/2 tions lead to Coriolis resonances and mixing between levels. A The (200) ∼ 0310 Coriolis interaction for SrOH has been 2Π 1/2 suggested previously [49]. Combination transitions requiring changes in multiple v quanta induced by the Coriolis inter- = 662 nm λ 2 actions have previously been observed in other polyatomic

= 663 nm molecules [69]. λ 1 (200) 2Σ To quantitatively estimate the vibrational transition moment ~ (100) 2Σ between (200) and (0310), we consider here the interactions X 2 that induce a strong transition dipole moment between (200) (000) Σ and (0310) in SrOH. By the symmetry of a linear molecule, anharmonic perturbations must be even in the bending normal coordinate Q2 and therefore can’t change v2 by an odd num- Figure 2. Internal quantum state preparation for SrOH via two-stage ber. Likewise, Coriolis interactions change v1 +v2 by an even optical pumping (OP). Using two OP laser beams (λ1 and λ2), a number at all orders of perturbation theory. Thus neither an- trapped SrOH sample can be prepared in a specific rotational quan- harmonic, nor Coriolis, effects alone can induce a transition tum level of the excited Sr-O stretching vibrational level (200). Spin- with 4v1 = 2 and 4v2 = 3. However, a combination of an- orbit splitting in the excited electronic state is indicated with both harmonic and Coriolis interactions lead to a relatively strong Π1/2 and Π3/2 levels shown. transition between (200) and (0310). Matrix elements for the Coriolis interaction couple v1 and v2, and may be found in [70]. Their strength is character- a two-stage optical pumping scheme from the ground vibra- ized by the Coriolis coefficient ζ21, which depends only on the tional level can populate (200) via two stages of excitation atomic masses and geometry of a molecule [71]. For SrOH, ˜2 to vibrationally excited levels of the A Π1/2 electronic state we find ζ21 = 0.98. (see Fig.2). In the first stage, molecules would be excited The vibrational potential energy for a linear polyatomic ˜2 to A Π1/2(100), which efficiently decays to X˜ (100). That molecule expanded in terms of the dimensionless normal co- p state, in turn, could be excited to A˜(200), which would like- ordinates qi = Qi 2πcωi/~ is wise preferentially decay to X˜ (200). Previous work on colli- sional quenching of the X˜ (100) state of SrOH at 2 K has al- 1 2 1 1 V/hc = ∑ωiqi + ∑φi jkqiq jqk + ∑ φi jklqiq jqkql +... ready demonstrated high-efficiency optical pumping into the 2 i 6 i jk 24 i jkl excited Sr-O stretching mode with a 660 nm external cavity (17) diode laser [68]. Thus, efficient rotational state preparation where φi jk and φi jkl are the cubic and quartic anharmonic in the (200) state can be achieved with two optical pumping force constants, respectively [72]. We use force constants up beams. to quartic order, computed from the potential energy surface An alternative transfer scheme from the ground vibrational (PES) calculation in [73]. As has been observed for CaOH ˜ ˜ 1 2 level to the excited X (200) state is to turn off the X (200) → [74], the term φ122q1q cannot be treated perturbatively due ˜2 6 2 A Π1/2 (100) repumping laser during the laser cooling pro- to vibrational near-resonances; we therefore directly diagonal- cess, thus leading to the rapid accumulation of molecules in ize the Hamiltonian including the full vibrational energy with ˜ the X (200) vibrational level. Each of the proposed methods anharmonic terms φi jk and φi jkl, as well as the Coriolis inter- appears highly feasible and the exact requirements of the fu- −1 action. Our numerical results give E100−000 = 515 cm and ture experiment will determine the preferred internal transfer −1 E0110−000 = 333 cm , agreeing with experimental observa- scheme. tions to better than 10%. As expected, the (200) and (0310) are found to be degenerate within the 10% estimated uncertainty of the ab initio energies. B. Transition strength Diagonalizing the Hamiltonian produces a set of vibrational eigenstates |ψ l i expanded in terms of the harmonic os- v1v2v3 For linear molecules the intensity of rovibrational transi- cillator basis, where the subscript labels the predominant ba- tions within the same electronic state is estimated as SJ0J00 = 2 4J sis component of the state. We then compute the transition |Mv0v00 | S 00 F (m), where Mv0v00 represents a purely vibra- J dipole moment as ∑v0,v00 hψ200|Mv0v00 |ψ0310i, where here Mv0v00 4J tional transition moment, SJ00 is the Hönl-London factor and gives the characteristic transition strength between hypotheti- F (m) is the Herman-Wallis term that compensates for errors cal pure harmonic oscillator states. Following the discussion in separation of vibration from rotation [52]. While for a in Sec.IVC, we estimate that Mv0v00 = 0.4 D for stretching purely harmonic oscillator only 4v = ±1 transitions are al- mode transitions in the harmonic oscillator basis, i.e. where lowed, inclusion of anharmonic terms in the molecular vi- ∆v1 = ±1 and ∆v2 = 0. Likewise, we estimate that Mv0v00 = 0.1 brational potential as well as high-order terms in the dipole D for bending mode transitions in the harmonic oscillator ba- 6

2 2 −20 sis, with ∆v2 = ±1 and ∆v1 = 0 (see [75]). shift is proportional to vthermal/c and will be 4ωRD ≈ 10 ω The resulting vibrational transition dipole moment for for an ultracold SrOH sample. (200) ↔ (0310) is estimated to be in the range 0.02 − 0.04 Blackbody radiation (BBR) can cause AC Stark shifts of D, depending slightly on the specific rotational transition con- molecular energy levels. In order to determine whether BBR- sidered due to the J-dependence of the Coriolis interaction. induced light shifts will cause an issue for the proposed mea- This compares favorably with other proposed measurements, surements we need to consider the differential BBR-shift for which typically rely on transition dipole moments of order the two ro-vibrational levels under consideration as well as the ≤ 0.01 D [36–39, 67]. experimentally viable value for the time-stability of the blackbody environment surrounding the molecular cloud. The frequency shift for each level under consideration is [56] C. Estimation of vibrational lifetime Z 3 2 BBR 4π ν Mi j The coherence time in the experiment will be limited by 4i = 3 ∑P dν . (20) 3ε0hc h exp(hν/kBT) − 1 ν − νi j the spontaneous vibrational lifetime of the (200) vibrational j ˜ ˜ state. Specifically, the decay rate X (200) → X (100) can be In order to estimate the magnitude of 4BBR, we can recast −7 3 2 estimated as A200−100 = 3.136×10 ω˜ M200−100 where ω˜ = Eq. 20 in terms of convenient experimental units, 522 is the energy splitting in cm-1 and the transition dipole moment M ≈ 0.4 is in Debye [52]. The dipole moment 200−100 3.136 × 10−7 ω˜ was calculated as [56] 4BBR = T˜ 3 M2 F i j , (21) i 2 ∑ i j ˜ 4π j T s dM M = ~ 200−100 (18) where T˜ is the temperature in cm-1, M is in Debye and F (y) 200−100 m ω dR red 1 R=Re is an integral function introduced by Farley and Wing [79] to evaluate the BBR-induced shift in the case of an E1 transi- where we used the approximate value for the slope of the tion. Since the BBR spectrum peaks around 600 cm-1, which dipole moment at the equilibrium separation of 3.17D/a0 happens to be close to vibrational transitions in SrOH, we estimated for the isoelectronic molecule SrF. The result- consider BBR-induced shifts due to vibrational transition res- ing lifetime is 1/A200−100 ≈ 140 ms. The black body in- onances. For room temperature, T˜ ≈ 200 cm-1, and using duced decay rate ΓBBR is further suppressed by a factor -1 ω˜ 1 ≈ 530 cm and previously estimated Mv00v0 ≈ 0.4 Debye 1/(exp(~ω1/(kBT)) − 1) ≈ 0.1 at room temperature [76]. (see Sec.IVC), we obtain 4BBR ∼ 1mHz. This is consistent with the estimations provided in Ref. [56] for other similar molecules. For the rotational transitions ω˜ i j T˜, we V. ESTIMATION OF SYSTEMATIC ERRORS obtain an asymptotic expression F (y) ' −π2y/3 [79] and BBR 4i . 1mHz. Here we show that several anticipated systematic errors can While the absolute magnitude of the BBR shifts seems be suppressed to below the target measurement precision, ow- to be large for a given ro-vibrational state, Vanhaecke and ing to the large enhancement factors and favorable molecular Dulieu pointed out that the differential dynamic BBR shift BBR BBR structure of SrOH. 4i − 4 j for molecular ro-vibrational transitions can be have a relative uncertainty of ∼ 10−13 −10−14 [56]. In particular, the vibrational dependence of the molecular dipole mo- A. Line broadening and shift ments for simple polyatomic molecules is on the order of part per hundred [80] and therefore potentially leads to a contri- As previously experimentally demonstrated with atomic bution to the differential BBR shift at the level of ∼ 10µHz, microwave clocks [77] and theoretically analyzed for a YbF which is on the order of the absolute statistical uncertainty molecular fountain [78], laser-cooled samples provide excel- for one day of experimental integration. Atomic clock ex- lent suppression of possible systematic errors in precision periments have characterized the magnitude of BBR shifts −4 measurement experiments. Doppler broadening is given by with a fractional uncertainty of . 5 × 10 [81]. We do not [52] anticipate any significant BBR anisotropies (like in trapped molecular ion experiments, for example [37]). Assuming re- r 2kT ln(2) alistically that BBR shifts can be characterized at the part ∆ω = 2ω (19) D mc2 per thousand level, even in the worst case scenario of estimated absolute shift 4BBR ∼ 1 mHz, the resulting fractional and will be suppressed at ultracold temperatures (∼ 50µK) uncertainty in the transition frequency measurement will be −10 BBR −15 to 4ωD ≈ 5 × 10 ω, which is 2 orders of magnitude lower δω /ω . 10 , thus not limiting the experimental preci- than for a 1 K sample of SrOH and below the natural linewidth sion at the level of δµ/µ ∼ 10−17 due to the large enhance- 1/Tc for ω200−0310 = 2π×1.2 GHz, illustrating one advantage ment factor Qµ 100. A recent work by Norrgard and co- of driving a transition between near-degenerate states to sup- workers describes a method to use molecules with optical cy- press systematic effects. The second order relativistic Doppler cling properties to perform quantum blackbody thermometry 7

−4 −5 with temperature sensitivity of σT /T ≈ 10 −10 [82]. Us- Parameter (200)(0310) ing such methods to perform in situ measurements of T BBR B [MHz] 7,384.788 [49] 7,429.631 [54] with trapped SrOH would allow further control over the BBR- BBR γ [MHz] 72.77 [49] 71.14 [54] induced systematics and enable δ4i 1µHz. q [MHz] 0 −12.484 [54] b [MHz] 1.713 [83] 1.713 [83] c [MHz] 1.673 [83] 1.673 [83] B. Field-insensitive transitions D [Debye] 1.900 [86] 1.900 [86]

To assess the sensitivity of our proposed approach to Table I. Molecular constants used to calculate electric and magnetic electric- and magnetic-field-induced systematic errors, we transition sensitivity compute the full energy level structure of the X˜ (200) and X˜ (0310) states up through the lowest 3 rotational levels in each state (N = 2 and N = 3, respectively). The Hamiltonian D have been previously measured for SrOH and are given takes the form in Tab.I with appropriate references for both the (200) and (0310) states. H = Hrot +HSR +H` +HFermi +Hdd +HStark +HZeeman, (22) As an example of this system’s robustness against sys- where Hrot is the rotational Hamiltonian, HSR is the spin- tematic errors, we will first consider transitions between the 00 0 rotation Hamiltonian, H` is the `-doubling Hamiltonian ap- N = 1 → N = 1 manifolds of the X˜ (200) → X˜ (0310) transi- 1 plicable to the (03 0) state, HFermi is the Fermi contact hyper- tion. fine interaction, H arises from the electron spin-nuclear spin dd In our numerical calculations, a reliable estimate of the en- dipole-dipole interaction, H is the Stark interaction, and Stark ergy shifts in state N requires explicit diagonalization of the H is the Zeeman interaction. Matrix elements for the Zeeman Hamiltonian including states up through N + 1. This can be rotational, spin-rotation, and `-doubling Hamiltonians may be understood as follows. At vanishing electric field, = 0, the found in [70] for both the bending and non-bending vibra- E dipole moment of every state is zero. In the (200) manifold, tional states. The hyperfine Hamiltonians are H = b I · S Fermi F the mixing of nearby rotational states at non-zero electric field and H = c(I S − I · S/3) [83], with matrix elements found dd z z induces a dipole moment in each sublevel. As a result, the in Ref. [84]. Likewise, the Stark interaction matrix ele- calculated induced dipole moments at low field are only valid ments may be found in Ref. [84]. Following [58], we use for state N if the basis includes the N + 1 state. Although H = g µ S· +g µ (S +S ) where S and Zeeman S B B l B xBx yBy x(y) Bx(y) the induced dipole moments in (0310) states are dominated are written in the molecular frame. The electron g-factor by mixing within `-doublets, rotational mixing occurs at the is constrained to its nominal value of g = 2.002319, and S same level as for the (200) manifold and therefore must also g = −γ/(2B) is given by the Curl identity. The relevant ma- l be considered to obtain an accurate estimate of the electric trix elements are available in Ref. [84], where we can use field sensitivity of a transition between the (200) and (0310) standard methods to transform from the lab frame to the B manifolds. In the calculations presented here, our basis spans molecule frame [85]. up to N = 2 in (200) and N = 3 in (0310), for a total of 156 While most of these Hamiltonians and matrix elements are states. readily available in the literature, it would be easy to over- look that the general form of the spin-rotation Hamiltonian is The measurement will be robust against systematic errors HSR = γS·(J −S−G), where γ is the spin-rotation constant, J related to electric (magnetic) field drifts if the transition has is the total angular momentum excluding nuclear spin and G a negligible difference between the ground and excited state is the angular momentum associated with the vibrational mo- electric (magnetic) dipole moments. We numerically diago- tion [70]. As a result, in the bending vibrational state, HSR nalize the full Hamiltonian in each vibrational manifold at a includes an interaction between the electron spin and vibra- variety of magnetic and electric fields and compute the lo- tional angular momentum along the molecular symmetry axis. cal dipole moments of each sublevel from the change in en- The spin-rotation splitting of a rotational state N in (0310) is ergy with respect to field strength. In considering the relative therefore given by dipole moments between two states, we restrict our attention to those whose overall transition strength is at least a non- γ 1 1 negligible fraction (e.g., 10%) of the strongest transition. ∆E(N) = 2N + 1 − − . (23) 2 N N + 1 See Fig.3 for the relative dipole moments of strong transitions among N00 = 1 → N0 = 1. The sharp vertical line in In a non-bending mode, the last two terms, which decrease Fig.3 arises from a resonance between opposite-parity states N with higher , are absent. In addition, these terms are some- in (0310) as the magnetic field is tuned. The thick, red transi- times neglected for spectroscopy of bending modes when the tions have the approximate composition high-N limit is appropriate. However, for the low-N states of interest here, all terms must be retained. The rotational constant B, spin-rotation constant γ, `- |(200),N00 = 1,J00 = 1/2,F00 = 0 − 1,M00 = 0i doubling constant q, Fermi contact coefficient bF , dipole- 1 0 0 0 0 dipole hyperfine coefficient c, and electric dipole moment → |(03 0),N = 1,J = 1/2,F = 0 − 1,M = 0i (24) 8

Transition Stark sensitivity at 1 mV/cm, N''=1→N'=1 Transition Zeeman sensitivity at 1 mV/cm, N''=1→N'=1 10 2

5 1

0 0

-1 -5

-2 [ Hz /( V / cm )] sensitivity field Electric

-10 [ MHz / gauss ] sensitivity field Magnetic 0 5 10 15 20 0 5 10 15 20 Magnetic field[gauss] Magnetic field[gauss]

Figure 3. Relative electric (left) and magnetic (right) dipole moments of strong transitions among N00 = 1 → N0 = 1. The thick, red lines show two transitions with low relative electric dipole moments and nearly zero common mode sensitivity to electric or magnetic fields around 6.40 G. and level, thus reducing systematic errors in common-mode resonance to the µHz level, which is below the frequency un- 00 00 00 00 |(200),N = 1,J = 3/2,F = 1 − 2,M = 0i certainty obtainable with one day of experimental integration. → |(0310),N0 = 1,J0 = 3/2,F0 = 1 − 2,M0 = 0i, (25) The electric dipole moment can be fine-tuned, and its sign can be reversed, with changes in magnetic field on the order of where a dash denotes an even superposition of different F 1−10 mG. states. We have identified additional favorable transitions in the In addition to having comparatively small individual rel- range of 0−20 G at fields around 5.94 G, 18.75 G, and 19.07 G ative dipole moments, these highlighted transitions can be for the N00 = 1 → N0 = 1 manifold. Furthermore, it is straight- made to have nearly exactly opposite sensitivities to both forward to find transitions between other rotational manifolds electric and magnetic fields. In particular, they have rela- with suppressed sensitivity to systematic errors. As an exam- tive g-factors of +0.1105 and −0.1099 at fields of 6.40 G ple, see Fig.4 for a pair of transitions in the N00 = 1 → N0 = 2 and 1 mV/cm, for a common-mode sensitivity to magnetic manifold with a nominal average g-factor of ∆g/2 ∼ 4×10−4 −4 fields characterized by ∆g/2 ∼ 3 × 10 . Therefore, simul- and average polarizability of −6µHz/(mV/cm)2. Once again, taneously measuring the resonance frequency of both transi- this magnetic moment is consistent with 0 at the level of exist- tions, and averaging the results, allows near complete elimi- ing spectroscopy and the average electric sensitivity of these nation of magnetic field-induced systematic errors. Although transitions can be fine-tuned and reversed with small adjust- many pairs of opposite-magnetic-sensitivity transitions ex- ments of the magnetic field. Comparably favorable transitions ist, it is typically the case that such pairs of transitions have have been found for the N00 = 0 → N0 = 1 rotational transition. large individual and common-mode electric field sensitivity at any particular magnetic field; thus simultaneously suppressed common-mode electric and magnetic relative dipole moments are non-trivial and must be found numerically. The VI. SENSING COSMIC FIELDS above-estimated common-mode magnetic dipole moment is smaller than the uncertainty arising from existing Zeeman The proposed measurement is predominantly sensitive to spectroscopy of SrOH. In particular, the rotational and nu- oscillation frequencies between the inverse of the total mea- clear g-factors, expected to be of order 10−3, have not yet been surement time (e.g., 1 day or 1 year) and the X˜ (200) decay measured in SrOH. With refined measurements of the Zeeman rate. We perform least-squares spectral analysis (LSSA) on structure, the optimal conditions to minimize the common- simulated data sets to quantify the projected sensitivity [87– mode sensitivity to magnetic fields could be fine-tuned. 89]. This method is closely related to the discrete Fourier In a similar manner, the two transitions considered transform but can be applied to the experimentally realistic sit- above have nearly opposite electric polarizabilities of −1081 uation in which data are not uniformly distributed in time, and Hz/(V/cm)2 and +1085 Hz/(V/cm)2, for an average polariz- accommodates inspection of arbitrary oscillation frequencies. ability of only 2µHz/(mV/cm)2 at a magnetic field of 6.40 G. We briefly summarize the LSSA approach here. For a discrete A transition in SrOH, or other co-trapped species, with sev- series of measurements, ω(ti), made at times {ti}, we fit the eral hundred times greater sensitivity to electric fields could data to a model ω(ti) = A j sin(2π f jti) + B j cos(2π f jt) + Cj, be used as a reference to actively stabilize the electric field where A j, B j, and Cj are fit parameters and f j is a possible os- over the small volume of the optical dipole trap to the mV/cm cillation frequency of the resonance. The estimated amplitude 9

Transition Stark sensitivity at 1 mV/cm, N''=1→N'=2 Transition Zeeman sensitivity at 1 mV/cm, N''=1→N'=2 4 2

2 1

0 0

-1 -2 -2 Hz /( V / cm )][ sensitivity field Electric

-4 [ MHz / gauss ] sensitivity field Magnetic 0 5 10 15 20 0 5 10 15 20 Magnetic field[gauss] Magnetic field[gauss]

Figure 4. Relative electric (left) and magnetic (right) dipole moments of strong transitions among N00 = 1 → N0 = 2. The thick, red lines show two transitions with low relative electric dipole moments and nearly zero common mode sensitivity to electric or magnetic ﬁelds around 12.75 G.

q 2 2 would generate a measurement signal-to-noise ratio (SNR) of of oscillation at frequency f j is then δωc( f j;δω) = A j + B j , 1 is then given by where δω is the true oscillation amplitude at f j. This procedure is repeated for each oscillation frequency f that is of j δωSNR=1( f j) δωc( f j;0) interest. = . (26) 1 Hz ( f ) For our simulation, we suppose that N = 106 trapped δωc j;1 Hz molecules are probed approximately every coherence time We repeat this procedure for the case of data interspersed Tc = 140 ms, with random delays of order ∼ 0.1Tc be- throughout one year, with one 24-hour series of measurements tween subsequent measurements. The single-measurement repeated weekly. In this case, the sensitivity at intermediate√ frequency sensitivity is assumed√ to be shot-noise limited, with and high frequencies improves by approximately 52 due to −1 statistical uncertainty ∆ω = (Tc N) [36, 55]. We first sim- the shot-noise limit, and the low-frequency noise cutoff is re- ulate the case of a single series of measurements, ω(ti), over duced to ∼30 nHz, set by the inverse of the total measurement 24 hours, with no assumed oscillation of the resonance fre- time of 1 year. quency. The inferred values of δωc( f j;0) > 0 arise from sta- Using the estimated enhancement factor Qµ ≈ −617 and tistical noise and allow an estimation of the noise floor of the transition frequency of 1.1 GHz, we find the fractional µ- measurement, ∆δω( f j). In the case of measurements over one variation at frequency f j corresponding to a signal-to-noise day, we find a noise floor of ∆δω( f j) ≈ 2π×3 µHz for f & 10 ratio of unity, (δµ/µ)SNR=1( f j) = δωSNR=1( f j)/(ωQµ). The µHz, as expected from the shot-noise limit. At low frequen- oscillation frequency f is related to the mass m of the new −2 5 2 cies, f j . 10 µHz, the sensitivity falls off as f j because the scalar particle by f = 2.42 × 10 (mc /neV) Hz [90]. inverse of the total measurement time, 24 hours, is ∼ 10 µHz The discussion above allows us to interpret the sensitivity and lower frequencies cannot be resolved from an offset in the of the measurement in terms of δµ/µ as a function of the mass mean of the resonance. mφ of a possible scalar dark matter particle φ. To go further To calculate the sensitivity to oscillation of the resonance at we must consider concrete models. As an example, we con- frequency f j, we then simulate a series of measurements with sider models of ultralight scalar particles with dilatonic in- d d a large oscillating resonance ω(t) = ω0 + δω( f j)sin(2π f jt + teractions, characterized by coupling constants me , g, and d φ j), where φ j ∈ [0,2π) is chosen randomly and we set mˆ , which arise from couplings of φ to electrons, gluons, and δω( f j) = 2π × 1 Hz. Each measurement of N molecules the symmetric combination of up and down quarks, respec- gives a measured value of ω equal to the mean of ω(t) over tively [18]. Assuming the new scalar particle comprises all of the 140 ms duration of the measurement, up to statistical dark matter, [18, 91] shot noise. At all sufficiently low oscillation frequencies, −1 f j . (2πTc) ∼ 1 Hz, the inferred amplitude of oscillation δµ = (dme − dg + MAdmˆ )κφ(t), (27) is accurate to excellent precision, δωc( f j;1 Hz) ≈ 1 Hz. At µ −1 large oscillation frequencies, f j (2πTc) , the sensitivity −13 −18 & where κφ0 = 6.4 × 10 (10 eV/mφ) with φ0 being the −1.4 falls off approximately as δωc( f j;1 Hz) ∝ f j because the amplitude of the time-dependent dark matter field φ(t) and average shift in the resonance frequency averages toward 0 MA = 0.037 quantifies the variation of the nucleon mass with over many oscillations. the quark mass in the case considered here of a transition di- The oscillation amplitude δωSNR=1( f j) at frequency f j that rectly sensitive to the proton-to-electron mass ratio [34]. 10

From Eq. 27 we can interpret the experimental sensitivity to 105 µ variation in terms of sensitivity to dme , dg, and dmˆ . Because |MA| < 1, the parameter space probed for dmˆ is less stringent than for dme and dg. The sensitivity to these parameters is shown as a function of mφ in Fig.5. For comparison, we also 100 show existing bounds on dme and dmˆ obtained from equivalence principle (EP) tests. The proposed measurement with SrOH would improve on H/Si the EP tests by up to about 8 orders of magnitude at the most EP limits sensitive frequency with 52 days of data, or over almost a 10-5 decade in mass range with only 24 hours of data. The largest Rb/Cs sensitivity to the coupling coefﬁcient Γµ between potential UDM coherent oscillations and proton-to-electron mass ratio in a one-day measurement occurs for dark matter particles -10 −20 −14 10 in the mass range mφ ∼ 5 × 10 eV to 1 × 10 eV, corresponding to oscillation periods of one day and the Nyquist 10-25 10-20 10-15 period 2Tc, respectively. If measurements are interspersed throughout a year, masses as low as 1 × 10−22 eV can be probed [90]. These mass ranges and the coupling coefﬁcients in the range shown in Fig.5 are already of interest to fun- Figure 5. Dashed lines depict sensitivity of the proposed measure- damental particle physics [92–94]. The use of quantum en- ments to select ultralight dark matter couplings, assuming single one- day measurement, or 52 one-day measurements interspersed weekly hanced metrology methods experimentally demonstrated for over a year. The low-frequency cusp corresponds to masses with microwave clocks can lead to further gains in sensitivity [95]. corresponding compton frequency at one day (year), below which −23 For mφ & 10 eV, our projected limits for 52 days of integra- sensitivity falls off rapidly, while the high-mass cusp corresponds to tion will improve existing experimental bounds from atomic the decay rate of X˜ (200). Solid blue (upper) and red (lower) lines spectroscopy over a 6-year time period [96, 97] by 4 orders indicate existing limits from the equivalence principle (EP) tests for of magnitude and will be complementary to future proposed dme and dmˆ terms, respectively [13]. Solid black line shows limits searches using atomic clocks as they probe different combina- on the dmˆ − dg coupling term from dual Rb/Cs cold atomic fountain tions of the coupling constants di [18]. at LNE-SYRTE [96, 97], while solid magenta line depicts a limit on dme from a comparison of a hydrogen maser with a crystalline sili- cone optical cavity [98]. VII. SUMMARY

We have considered the search for ultralight dark matter us- trapping [100, 101]. Previous high-resolution vibrational ing precision microwave spectroscopy of the laser-cooled tri- spectroscopy [101–103] of CaOH predicts a transition energy atomic radical SrOH. The rovibronic spectrum of SrOH in the of only ∼ 0.1cm−1 for the Q(N = 4) transition of X˜ (0600) → ground electronic state has been analyzed, and the enhance- X˜ (1440), with an associated enhancement factor estimated to ment factors Qµ are calculated for different rotational transi- be |Qµ| ∼ 500. These states are subject to signiﬁcant anhar- tions in the (200) ↔ 0310 vibrational band. With predicted monic contributions and Coriolis resonances, and differ by

Qµ 10 for multiple rovibrational transitions, as well as only 5 vibrational quanta like the states of interest for SrOH; highly diagonal Franck-Condon factors in the X˜ ↔ A˜ elec- it is therefore reasonable to expect a similarly strong transi- tronic excitation band, laser-cooled SrOH provides a viable tion moment as analyzed above. Thus further spectroscopy molecular platform for achieving ∼ 10−17 uncertainty in δµ/µ and characterization of these states may reveal an alternative with 1 day of integration and has the potential to signiﬁcantly route to probe δµ/µ via precision measurement of rovibra- improve on the previous limit on δµ/µ from molecular spec- tional transitions in triatomic MOH molecules. troscopy [33]. The higher density of rovibrational states provided by the Looking for signatures of high-energy physics in low- mechanical motion of MOR molecules with more complex energy spectroscopy experiments with laser-cooled SrOH has ligands could result in similar degeneracies as analyzed here the potential to complement other experimental efforts to un- but with even larger enhancement factors Qµ, enabling ac- cover the quantum mechanical nature of the dark sector of cess to a new UDM-coupling range by probing δµ/µ frac- −18 the universe [13, 16]. Furthermore, while SrOH is one of tional uncertainty in the . 10 regime. For example, re- the simplest examples of monovalent metal alkoxides (MOR) cently laser cooled MOR-type symmetric top molecule cal- that have been previously identiﬁed as suitable for direct laser cium monomethoxide CaOCH3 [104] possesses two nearly cooling and trapping [99], degeneracies between vibrational degenerate vibrational modes arising from the mechanical -1 states of different character are ubiquitous among polyatomic motion of the CH3 group (ωvib ∼ 1,450 cm ). Previous ab molecules. For example, CaOH is another triatomic molecule initio calculations predict that the CH3 umbrella (a1 symme- which has since been laser cooled in a cryogenic beam and is try) and scissoring (e symmetry) motions are less than 20 actively being pursued for three-dimensional magneto-optical cm-1 apart [105, 106], which can be further reduced to ∼ 1 11 cm-1 by driving perpendicular rovibrational transitions with version of the manuscript. We would also like to acknowl- K00 = 1 → K0 = 2. While further experimental measurements edge insightful discussions with J. Weinstein during the early are needed to identify the contributions from anharmonic parts stages of this work and thank B. Augenbraun for bringing to of the potential in order to accurately predict the enhancement our attention the feasibility of ro-vibronic near-degeneracies factors Qµ, the presence of new rotational degrees of freedom in CaOH. We are grateful to J. Kłos and S. Kotochigova for compared to linear molecules enables precise “tuning” of the sharing the vibrational potential of SrOH. separation between near-degenerate levels. I.K. and Z.L. contributed equally to this work.

ACKNOWLEDGMENTS Appendix A: Estimation of SrOH molecular constants

This work has been funded by the AFOSR Grant No. In the past, extensive molecular spectroscopy has been per- FA9550-15-1-0446 and NSF Grant No. PHY-1505961. We formed on SrOH with many vibrational and rotational param- would like to thank M. Safronova for encouraging us to pursue eters precisely measured [49]. In a ground electronic state, the topic of dark matter effects in the spectra of laser-cooled vibrational energy levels of a linear triatomic molecule like polyatomic molecules and for a critical reading of the initial SrOH are given by [52]

3 3 l di di dk 2 G v1v2v3 = ∑ ωi vi + + ∑ ∑ xik vi + vk + + g22l (A1) i=1 2 i=1 k≥i 2 2

where di = 1 for non-degenerate stretching vibrations (v1 and 4E0220−0200 = 4g22 = 30.233 (A6) v3) and di = 2 for the doubly-degenerate bending mode v2. For SrOH and other similar molecules, the low-lying vibra- E0110−000 = ω2 + 3x22 + g22 = 363.687 (A7) tional motions are mostly harmonic and therefore ωi xii, xik for i 6= k. Therefore, SrOH vibrational levels of experimental relevance are approximated by the following expression: E0200−000 = 2ω2 + 8x22 = 703.288. (A8)

1 1 E ' ω v + + ω (v + 1) + ω v + (A2) Appendix B: Normal modes of a linear triatomic molecule v1v2v3 1 1 2 2 2 3 3 2 In order to determine the dependence of vibrational fre- 2 2 quencies of SrOH on the proton-to-electron mass ratio µ, we 1 2 1 2 +x11 v1 + + x22 (v2 + 1) + x33 v3 + + g22l . perform the normal mode analysis using the GF matrix for- 2 2 malism [107]. The kinetic-energy-related matrix G for a lin- Using Eq. A2 as well as the measured energies of the (100), ear triatomic molecule is given by (200), 0110, 0200 and 0220 states [49], we can esti-   µ1 + µ2 −µ3 0 mate all of the necessary harmonic (ω1 and ω2) as well as G =  −µ3 µ2 + µ2 0 , (B1) anharmonic (x11, x22 and g22) constants. It is computationally 0 0 G convenient to reference all of the excited vibrational levels rel- 33 ative to the ground vibrational level where following a common convention in the literature we use the notation µ ≡ 1/m , µ = 1/m and µ ≡ 1/m while ω1 ω3 x11 x33 1 Sr 2 H 3 O E000 = + ω2 + + + x22 + . (A3) 2 2 4 4 2 r32 r31 (r31 + r32) -1 G33 = µ1 + µ2 + µ3 , (B2) The estimated vibrational constants (in cm ) are ω1 = r31 r32 r31r32 531.900, x11 = −2.455, ω2 = 369.584, x22 = −4.485 and g = 7.558. With these extracted constants and using Eq. A2 which also has units of 1/[mass]. The diagonal force constant 22 matrix is given by for vibrational levels of SrOH, we predict positions of E100, -1 E , E 1 , E 0 and E 2 to 0.002 cm , which corresponds   200 01 0 02 0 02 0 F11 0 0 to 0.06 GHz. In particular, we have the following expressions F =  0 F22 0 . (B3) (in units of cm-1): 0 0 F33

E100−000 = ω1 + 2x11 = 526.991 (A4) 2 Solving for eigenvalues of GF and setting them equal to ωi , we can ﬁnd an expression for the harmonic vibrational fre- 4E200−100 = ω1 + 4x11 = 522.082 (A5) quencies in terms of atomic masses: 12

1 2 2 2 2 2 2 2 ω1,3 = F11 (µ1 + µ3) + F22 (µ2 + µ3) ∓ F11 (µ1 + µ3) + F22 (µ2 + µ3) + 4F11F22µ33 − 2F11F22 (µ1 + µ3)(µ2 + µ3)

2 ! 2 r32 r31 (r31 + r32) constant for the bending motion (see Eq. B3) and µbend ∝ mp ω2 = F33 µ1 + µ2 + µ3 (B4) is the reduced mass of the bending motion. r31 r32 r31r32 The anharmonic constants x11, x22 and g22 for SrOH can be expressed in terms of the force constants and vibrational where ω1, ω2 and ω3 refer to the harmonic vibrational frequencies for Sr-O stretching, bending and O-H stretching frequencies as [72] modes, respectively. Notice that since the binding energy of 2 2 2 4 2 1 1 2 8ω1 − 3ωi the nuclei is Eb ∼ ka = mee / in a molecule and thus the 0 ~ x11 = φ1111 − ∑φ11i 2 2, (C3) force constant k is proportional to the electron mass me [32], 16 16 i ωi 4ω1 − ωi the calculated vibrational frequencies are all proportional to r 2 − 2 me −1/2 1 1 2 8ω2 3ωi ωi ∝ = µ . (B5) x22 = φ2222 − ∑φi22 2 2, (C4) mp 16 16 i ωi 4ω2 − ωi

The stretch-stretch coupling constant F13 has been ignored in our calculations since it is less than 1% of the corresponding 1 1 2 ωi g22 = − φ2222 − ∑φi22 2 2 . (C5) F11 force constant and the use of the diagonal force matrix F 48 16 i 4ω2 − ωi has proven reasonably accurate in the previous work on SrOH [108]. The Morse potential provides a good approximation to bond-stretching motions of linear polyatomic molecules with −1 x11, x33 ∝ µ . Without loss of generality, consider me ﬁxed Appendix C: Anharmonic vibrations of triatomic molecules and, therefore, change in µ corresponds to change in mp [38]. From the dimensionality comparison of Eq. C3, C4 and C5 −1 Calculated spectroscopic constants for SrOH indicate that we conclude that x22, g22 ∝ µ . there is a small anharmonic contribution to stretching and bending molecular vibrations as can be seen above. Exact Appendix D: Extensions of proposed work description of the vibrational motion of polyatomic molecules requires inclusion of the anharmonic terms in the molecular potential. A Morse potential of the form [47] 1. Isotopic substitution

h i2 Vibrational frequencies of the normal modes in polyatomic E = E 1 − e−a(R−Re) (C1) Morse b molecules depend on the constituent atomic isotopes. Stron- tium has four stable isotopes with atomic masses 88, 86, 87 provides a good approximation for the anharmonic vibrational and 84 and natural abundances of 82.58%, 9.86%, 7% and potential of a diatomic molecule. It can be shown that the vi- 0.56%, respectively. Additionally, a deuterated version of the brational energy levels for a diatomic molecule take the form molecule SrOD has been previously experimentally analyzed [47] [111]. While for a diatomic molecule the dependence of the µ 2 2 2 molecular vibrational constants on the reduced mass red is 1 ~ ω0 1 −1/2 −1 Ev = ~ω0 v + − v + (C2) relatively simple, ω ∝ µred and x ∝ µred, even for a linear 2 4Eb 2 triatomic molecule the change in harmonic and anharmonic | {z } x vibrational constants as a function of isotopic substitution is more complex, as discussed above. While the focus of this −1/2 where the µ constant dependence manifests as ω0 ∝ µ for paper is on the most abundant 88Sr16O1H isotope, potentially −1 the harmonic and x ∝ µ for the anharmonic constant. Con- other SrOH isotopes could be useful for µ variation experi- tinuing to treat me as fixed without loss of generality, we note ments as well. e2 that the binding energy Eb ∼ Eel ∼ , where a0 is the Bohr a0 radius, does not directly depend on the proton mass mp and is therefore independent of µ [32]. 2. “Frozen” SrOH For a polyatomic molecule, local bond stretching vibrations like Sr↔O and O↔H can also be effectively treated as Morse In order to observe spectral signatures of the resonant ab- −1/2 oscillators [109] and therefore ω1, ω3 ∝ µ and x11, x33 ∝ sorption of bosonic dark matter previous proposals consid- µ−1. For bending vibrations of linear triatomic molecules like ered using a pressurized gas container at room temperature SrOH it can also be analytically shown [110] that vibrational with H ,O , CO, N , HCl or I [112] or a cryogenic buffer- p 2 2 2 2 levels become ~(v + 1) f /µbend where f ∝ me is the force gas-cooled sample of O2 molecules [113]. Alternatively, one 13 could consider using SrOH molecules embedded in a cryo- of using diatomic molecules embedded in a solid inert-gas genic noble-gas matrix. High atomic densities of order 1017 matrix has been proposed for performing EDM experiments cm-3 have been demonstrated with spin coherence times ap- with projected ∼ 10−37 e · cm sensitivity [115]. However, the proaching ∼ 1 s under some conditions [114]. Laser spec- approach with frozen polyatomic molecules for dark matter troscopy of the macroscopic sample of “frozen” SrOH could searches does not require the application of MV/cm external allow probing ALP masses in the µeV and meV range for electric fields for molecular orientation in the lab frame, thus dark-matter induced rotational and vibrational transitions, re- significantly simplifying experimental design. A more exten- spectively. We would like to point out that a similar approach sive analysis of this approach is beyond the scope of this work.

[1] V. A. Mitsou, in J. Phys. Conf. Ser., Vol. 651 (IOP Publishing, ers, and E. Peik, Phys. Rev. Lett. 113, 210802 (2014). 2015) p. 012023. [28] T. Rosenband, D. Hume, P. Schmidt, C.-W. Chou, A. Brusch, [2] L. Baudis, Phys. Dark Universe 1, 94 (2012). L. Lorini, W. Oskay, R. E. Drullinger, T. M. Fortier, J. Stal- [3] R. Agnese, T. Aramaki, I. J. Arnquist, W. Baker, D. Bal- naker, et al., Science 319, 1808 (2008). akishiyeva, S. Banik, D. Barker, R. B. Thakur, D. Bauer, [29] N. Leefer, C. Weber, A. Cingöz, J. Torgerson, and D. Budker, T. Binder, et al., Phys. Rev. Lett. 120, 061802 (2018). Phys. Rev. Lett. 111, 060801 (2013). [4] A. Tan, M. Xiao, X. Cui, X. Chen, Y. Chen, D. Fang, C. Fu, [30] B. M. Roberts, G. Blewitt, C. Dailey, M. Murphy, K. Giboni, F. Giuliani, H. Gong, et al., Phys. Rev. Lett. 117, M. Pospelov, A. Rollings, J. Sherman, W. Williams, and 121303 (2016). A. Derevianko, Nat. Commun. 8, 1195 (2017). [5] D. Akerib, S. Alsum, H. Araújo, X. Bai, A. Bailey, J. Balajthy, [31] P. Wcislo, P. Morzynski, M. Bober, A. Cygan, D. Lisak, P. Beltrame, E. Bernard, A. Bernstein, T. Biesiadzinski, et al., R. Ciurylo, and M. Zawada, Nat. Astron. 1, 1 (2016). Phys. Rev. Lett. 118, 021303 (2017). [32] D. DeMille, Phys. Today 68, 34 (2015). [6] L. Baudis, Phys. Dark Universe 4, 50 (2014). [33] A. Shelkovnikov, R. J. Butcher, C. Chardonnet, and A. Amy- [7] CMS Collaboration, J. High Energy Phys. 2018, 27 (2018). Klein, Phys. Rev. Lett. 100, 150801 (2008). [8] CMS Collaboration, arXiv:1712.08501 [hep-ex] (2017). [34] V. Flambaum, D. B. Leinweber, A. W. Thomas, and R. D. [9] W. B. Cairncross, D. N. Gresh, M. Grau, K. C. Cossel, T. S. Young, Phys. Rev. D 69, 115006 (2004). Roussy, Y. Ni, Y. Zhou, J. Ye, and E. A. Cornell, Phys. Rev. [35] X. Calmet and H. Fritzsch, Eur. Phys. J. C 24, 639 (2002). Lett. 119, 153001 (2017). [36] D. Hanneke, R. Carollo, and D. Lane, Phys. Rev. A 94, [10] ACME Collaboration, Science 343, 269 (2014). 050101 (2016). [11] P. W. Graham, D. E. Kaplan, and S. Rajendran, Phys. Rev. [37] M. G. Kokish, P. R. Stollenwerk, M. Kajita, and B. C. Odom, Lett. 115, 221801 (2015). Phys. Rev. A 98, 28 (2018). [12] Y. V. Stadnik and V. V. Flambaum, Mod. Phys. Lett. A 32, [38] D. DeMille, S. Sainis, J. Sage, T. Bergeman, S. Kotochigova, 1740004 (2017). and E. Tiesinga, Phys. Rev. Lett. 100, 043202 (2008). [13] K. Van Tilburg, N. Leefer, L. Bougas, and D. Budker, Phys. [39] V. Flambaum and M. Kozlov, Phys. Rev. Lett. 99, 150801 Rev. Lett. 115, 011802 (2015). (2007). [14] I. G. Irastorza and J. Redondo, Prog. Part. Nucl. Phys. 102, 89 [40] T. Zelevinsky, S. Kotochigova, and J. Ye, Phys. Rev. Lett. 100, (2018). 043201 (2008). [15] A. Berlin and N. Blinov, Phys. Rev. Lett. 120, 021801 (2018). [41] M. Kozlov, Phys. Rev. A 87, 032104 (2013). [16] D. Budker, P. W. Graham, M. Ledbetter, S. Rajendran, and [42] P. Jansen, L.-H. Xu, I. Kleiner, W. Ubachs, and H. L. Bethlem, A. O. Sushkov, Phys. Rev. X 4, 021030 (2014). Phys. Rev. Lett. 106, 100801 (2011). [17] F. D’Eramo, L. J. Hall, and D. Pappadopulo, J. High Energy [43] V. V. Ilyushin, J. Mol. Spectrosc. 300, 86 (2014). Phys. 2014, 108 (2014). [44] A. Owens, S. Yurchenko, W. Thiel, and V. Špirko, Phys. Rev. [18] A. Arvanitaki, J. Huang, and K. Van Tilburg, Phys. Rev. D 91, A 93, 052506 (2016). 015015 (2015). [45] J. Bagdonaite, P. Jansen, C. Henkel, H. L. Bethlem, K. M. [19] Y. Stadnik and V. Flambaum, Phys. Rev. Lett. 115, 201301 Menten, and W. Ubachs, Science 339, 46 (2013). (2015). [46] I. Kozyryev, L. Baum, K. Matsuda, B. L. Augenbraun, L. An- [20] A. H. Guth, M. P. Hertzberg, and C. Prescod-Weinstein, Phys. deregg, A. P. Sedlack, and J. M. Doyle, Phys. Rev. Lett. 118, Rev. D 92, 103513 (2015). 173201 (2017). [21] B. Roberts and A. Derevianko, arXiv:1803.00617 (2018). [47] W. Demtroeder, Molecular Physics: Theoretical Principles [22] M. G. Kozlov and S. A. Levshakov, Ann. Phys. (Berl.) 525, and Experimental Methods (Wiley-VCH, 2005). 452 (2013). [48] G. Herzberg, Molecular spectra and molecular structure. Vol. [23] P. Jansen, H. L. Bethlem, and W. Ubachs, J. Chem. Phys. 140, 3: Electronic spectra and electronic structure of polyatomic 010901 (2014). molecules (New York: Van Nostrand, Reinhold, 1966, 1966). [24] V. A. Dzuba, V. V. Flambaum, and S. Schiller, Phys. Rev. A [49] P. I. Presunka and J. A. Coxon, Chem. Phys. 190, 97 (1995). 98, 022501 (2018). [50] C. Brazier and P. Bernath, J. Mol. Spectrosc. 114, 163 (1985). [25] M. S. Safronova, S. G. Porsev, C. Sanner, and J. Ye, Phys. [51] D. Fletcher, M. Anderson, W. Barclay Jr, and L. Ziurys,J. Rev. Lett. 120, 173001 (2018). Chem. Phys. 102, 4334 (1995). [26] R. Godun, P. Nisbet-Jones, J. Jones, S. King, L. Johnson, [52] P. F. Bernath, Spectra of Atoms and Molecules (New York: H. Margolis, K. Szymaniec, S. Lea, K. Bongs, and P. Gill, Oxford University Press, 2005). Phys. Rev. Lett. 113, 210801 (2014). [53] S. Levshakov, M. Kozlov, and D. Reimers, Astrophys. J. 738, [27] N. Huntemann, B. Lipphardt, C. Tamm, V. Gerginov, S. Wey- 26 (2011). 14

[54] D. Fletcher, M. Anderson, W. Barclay Jr, and L. Ziurys,J. [85] J. M. Brown and A. Carrington, Rotational spectroscopy of Chem. Phys. 102, 4334 (1995). diatomic molecules (Cambridge Univ. Press, 2003). [55] A. D. Ludlow, M. M. Boyd, J. Ye, E. Peik, and P. O. Schmidt, [86] T. Steimle, D. Fletcher, K. Jung, and C. Scurlock, J. Chem. Rev. Mod. Phys. 87, 637 (2015). Phys. 96, 2556 (1992). [56] N. Vanhaecke and O. Dulieu, Mol. Phys. 105, 1723 (2007). [87] A. Cumming, G. W. Marcy, and R. P. Butler, The Astrophys- [57] I. Kozyryev, Laser cooling and inelastic collisions of the poly- ical Journal 526, 890 (1999). atomic radical SrOH, Ph.D. thesis, Harvard University (2017). [88] A. Cumming, Monthly Notices of the Royal Astronomical So- [58] D.-T. Nguyen, T. C. Steimle, I. Kozyryev, M. Huang, and ciety 354, 1165 (2004). A. B. McCoy, J. Mol. Spectrosc. 347 (2018). [89] C. Abel, N. Ayres, G. Ban, G. Bison, K. Bodek, V. Bondar, [59] I. Kozyryev, L. Baum, L. Aldridge, P. Yu, E. E. Eyler, and M. Daum, M. Fairbairn, V. Flambaum, P. Geltenbort, et al., J. M. Doyle, Phys. Rev. Lett. 120, 063205 (2018). Phys. Rev. X 7, 041034 (2017). [60] M. Morita, J. Kłos, A. A. Buchachenko, and T. V. Tscherbul, [90] A. Derevianko, Phys. Rev. A 97, 042506 (2018). Phys. Rev. A 95, 063421 (2017). [91] K. V. Tilburg, Searches for Light Scalar Dark Matter, Ph.D. [61] L. Anderegg, B. L. Augenbraun, E. Chae, B. Hemmerling, thesis, Stanford University (2016). N. R. Hutzler, A. Ravi, A. Collopy, J. Ye, W. Ketterle, and [92] W. Hu, R. Barkana, and A. Gruzinov, Phys. Rev. Lett. 85, J. M. Doyle, Phys. Rev. Lett. 119, 103201 (2017). 1158 (2000). [62] L. Anderegg, B. L. Augenbraun, Y. Bao, S. Burchesky, L. W. [93] D. J. Marsh and A.-R. Pop, Mon. Not. R. Astron. Soc. 451, Cheuk, W. Ketterle, and J. M. Doyle, Nat. Phys. 14, 890 2479 (2015). (2018). [94] V. Lora, J. Magana, A. Bernal, F. Sánchez-Salcedo, and [63] C. Cheng, A. P. Van Der Poel, P. Jansen, M. Quintero-Pérez, E. Grebel, J. Cosmol. and Astropart. Phys. 2012, 011 (2012). T. E. Wall, W. Ubachs, and H. L. Bethlem, Phys. Rev. Lett. [95] O. Hosten, N. J. Engelsen, R. Krishnakumar, and M. A. Ka- 117, 253201 (2016). sevich, Nature 529, 505 (2016). [64] N. Davidson, H. J. Lee, C. S. Adams, M. Kasevich, and [96] A. Hees, J. Guéna, M. Abgrall, S. Bize, and P. Wolf, Phys. S. Chu, Phys. Rev. Lett. 74, 1311 (1995). Rev. Lett. 117, 061301 (2016). [65] R. Parker, M. Dietrich, M. Kalita, N. Lemke, K. Bailey, [97] A. Hees, O. Minazzoli, E. Savalle, Y. V. Stadnik, and P. Wolf, M. Bishof, J. Greene, R. Holt, W. Korsch, Z.-T. Lu, et al., Phys. Rev. D 98, 064051 (2018). Phys. Rev. Lett. 114, 233002 (2015). [98] C. Kennedy, E. Oelker, J. Robinson, T. Bothwell, D. Kedar, [66] W. Li, Y. Du, H. Li, and Z. Lu, AIP Adv. 8, 095311 (2018). W. Milner, G. Marti, A. Derevianko, and J. Ye, [67] J. Kobayashi, A. Ogino, and S. Inouye, Nat. Commun. 10, arXiv:2008.08773 (2020). 3771 (2019). [99] I. Kozyryev, L. Baum, K. Matsuda, and J. M. Doyle, [68] I. Kozyryev, L. Baum, K. Matsuda, P. Olson, B. Hemmerling, ChemPhysChem 17, 3641 (2016). and J. M. Doyle, New J. Phys. 17, 045003 (2015). [100] L. Baum, N. B. Vilas, C. Hallas, B. L. Augenbraun, S. Raval, [69] D. B. Moss, Z. Duan, M. P. Jacobson, J. P. O’Brien, and R. W. D. Mitra, and J. M. Doyle, Physical Review Letters 124, Field, J. Mol. Spectrosc. 199, 265 (2000). 133201 (2020). [70] A. J. Merer and J. M. Allegretti, Canadian Journal of Physics [101] L. Baum, N. B. Vilas, C. Hallas, B. L. Augenbraun, S. Raval, 49, 2859 (1971). D. Mitra, and J. M. Doyle, arXiv:2006.01769 (2020). [71] J. H. Meal and S. R. Polo, The Journal of Chemical Physics [102] J. A. Coxon, M. Li, and P. I. Presunka, Molecular Physics 76, 24, 1119 (1956). 1463 (1992). [72] W. D. Allen, Y. Yamaguchi, A. G. Császár, D. A. Clabo Jr, [103] M. Li and J. A. Coxon, The Journal of Chemical Physics 102, R. B. Remington, and H. F. Schaefer III, Chemical Physics 2663 (1995). 145, 427 (1990). [104] D. Mitra, N. B. Vilas, C. Hallas, L. Anderegg, B. L. Augen- [73] M. Li, J. Kłos, A. Petrov, and S. Kotochigova, Commun. Phys. braun, L. Baum, C. Miller, S. Raval, and J. M. Doyle, Science 2, 148 (2019). 369, 1366 (2020). [74] J. Koput and K. A. Peterson, Journal of Physical Chemistry A [105] I. Kozyryev, T. C. Steimle, P. Yu, D.-T. Nguyen, and J. M. 106, 9595 (2002). Doyle, New J. Phys. 21 (2019). [75] I. Kozyryev and N. R. Hutzler, Phys. Rev. Lett. 119, 133002 [106] A. C. Paul, K. Sharma, M. A. Reza, H. Telfah, T. A. Miller, (2017). and J. Liu, J. Chem. Phys. 151, 134303 (2019). [76] S. Y. Buhmann, M. Tarbutt, S. Scheel, and E. Hinds, Physical [107] E. B. Wilson, J. C. Decius, and P. C. Cross, Molecular vibra- Review A 78, 052901 (2008). tions: the theory of infrared and Raman vibrational spectra [77] A. Bauch, Meas. Sci. Technol. 14, 1159 (2003). (Courier Corporation, 1955). [78] M. Tarbutt, B. Sauer, J. Hudson, and E. Hinds, New J. Phys. [108] M. Oberlander, Laser excited ﬂuorescence studies of reactions 15, 053034 (2013). of group 2 metals with oxygen containing molecules, Ph.D. [79] J. W. Farley and W. H. Wing, Phys. Rev. A 23, 2397 (1981). thesis, Ohio State University (1995). [80] A. S. Sharipov, B. I. Loukhovitski, and A. M. Starik, J. Phys. [109] H. Lefebvre-Brion and R. W. Field, The Spectra and Dynamics B 50, 165101 (2017). of Diatomic Molecules (San Diego: Elsevier Academic Press, [81] W. McGrew, X. Zhang, R. Fasano, S. Schäffer, K. Beloy, 2004). D. Nicolodi, R. Brown, N. Hinkley, G. Milani, M. Schioppo, [110] T. Hirano, U. Nagashima, and P. Jensen, J. Mol. Spectrosc. et al., Nature 564, 87 (2018). 343, 54 (2018). [82] E. B. Norrgard, S. P. Eckel, C. L. Holloway, and E. L. Shirley, [111] M. Anderson, W. Barclay Jr, and L. Ziurys, Chemical Physics arXiv:2011.06643 (2020). Letters 196, 166 (1992). [83] D. Fletcher, K. Jung, C. Scurlock, and T. Steimle, J. Chem. [112] A. Arvanitaki, S. Dimopoulos, and K. Van Tilburg, Phys. Rev. Phys. 98, 1837 (1993). X 8, 041001 (2018). [84] E. Hirota, High-Resolution Spectroscopy of Transient [113] L. Santamaria, C. Braggio, G. Carugno, V. Di Sarno, P. Mad- Molecules (Springer-Verlag Berlin Heidelberg, 1985). daloni, and G. Ruoso, New Journal of Physics 17, 113025 15

(2015). [115] A. Vutha, M. Horbatsch, and E. A. Hessels, Atoms 6, 3 [114] S. Upadhyay, A. N. Kanagin, C. Hartzell, T. Christy, W. P. (2018). Arnott, T. Momose, D. Patterson, and J. D. Weinstein, Phys. Rev. Lett. 117, 175301 (2016).